SignalCaster is built for unattended broadcasting. When the connection to your streaming server drops — for any reason — SignalCaster automatically attempts to reconnect and resume the stream without you needing to do anything. Here’s how it all works.
If the TCP connection to the server closes unexpectedly while you’re streaming, SignalCaster immediately attempts to reconnect. The first retry happens after a short delay; subsequent retries use increasing backoff intervals. The reconnect attempt count and each retry are logged in the Activity Log. Once the connection is re-established, streaming resumes automatically from where it left off (audio continues flowing — there’s no gap in the encoder).
Most streaming services (YouTube, Twitch, Mixcloud) handle reconnects gracefully. Your live stream stays active from the viewer’s perspective, with at most a brief buffering pause during the reconnect window.
A dropped TCP connection is the obvious failure mode — but sometimes a connection looks alive (TCP is established) while the RTMP session has quietly stalled. SignalCaster’s RTMP health monitor detects this "zombie connection" scenario with two complementary mechanisms:
Every 10 seconds SignalCaster sends an RTMP Ping Request to the server and waits for a Pong response. The round-trip time is logged. If the server doesn’t respond to a ping within a timeout window, SignalCaster treats the session as dead and triggers a reconnect.
SignalCaster also tracks whether the server is acknowledging the bytes it’s receiving. If you’ve sent more than 500 KB since the last server acknowledgement and a significant time has passed, the session is flagged as stalled and a reconnect is triggered.
Ping latency is measured with each Ping/Pong cycle. If SignalCaster sees three or more pings with latency above 500 ms within a two-minute sliding window, it treats this as a deteriorating connection and proactively reconnects before the session fully stalls. This avoids the frustrating "stream is technically connected but viewers are buffering constantly" situation.
Single spikes above 2,000 ms trigger an immediate reconnect without waiting for the trend threshold.
When latency is elevated, the ping interval drops from 10 seconds to 3 seconds so SignalCaster can react faster to a deteriorating connection. It returns to 10 seconds once latency normalizes.
Some services (notably Mixcloud and YouTube) can silently drop a stream after a very long continuous session — not due to any network problem, just the server ending the session on its side after many hours. SignalCaster tracks how long you’ve been continuously streaming and proactively reconnects before that window expires, keeping your stream live without any viewer-facing interruption.
Reconnect handles transient network issues and server-side session endings. It will not help if your stream key has expired, your account has been suspended, or your upload bandwidth is permanently saturated. If reconnect attempts are failing repeatedly, check the Activity Log for the specific error the server is returning.